A processor that evaluates a machine learning models built with TensorFlow and stored in Protocol Buffers binary format.
The TensorFlow processor leverages internally TensorFlow Java API library.
To learn more about this application and the supported properties, please review the following details.
To learn how to customize the processor for better fit different TensorFlow models, please review the
the Object Detection
, Pose Estimation
, Twitter Sentiment
and Image Recognition
processors below.
Object Detection Processor - The Object Detection processor provides out-of-the-box support for the TensorFlow Object Detection API. It allows for real-time localization and identification of multiple objects in a single image or image stream. The Object Detection processor uses one of the pre-trained object detection models and corresponding object labels. spring-cloud-starter-stream-processor-object-detection/README.adoc
Video Tutorial: Real-time Object-Detection With Spring Cloud Stream
Pose Estimation Processor
The Pose Estimation is a real-time, multi-person body pose estimation processor for detecting human figures in images and video. Used to determine, where different body parts are located in an image an how they spatially relate to each other. Processor is inspired and based on - Realtime Multi-Person 2D Pose Estimation using Part Affinity Fields - OpenPose |
Twitter Sentiment Processor - Extends the tensorflow-processor
for real-time twitter message sentiment analysis.
Image Recognition Processor - Real-time image recognition and classification processor. Based on InceptionV3 model spring-cloud-starter-stream-processor-image-recognition/README.adoc
Video tutorial : Real-time image recognition